Image segmentation is the key technology of image analysis ,pattern recognition and other fields.In order to reduce the impact of factors such as light to color and enhance the quality of image segmentation,a color image segmentation algorithm base on the Grid SearchSupport Vector Machine(GSSVM) is proposed.Firstly, the RGB and HSI components were extracted from image.Then components combined into the feature space of the sample for the training of SVM.The parameters of the SVM were optimized through the grid search method when training.Finally, aftertraining GSSVM segmented the color image.Experience has shown that the algorithm can find the global optimal solution within the space of the grid. The segmentation accuracy of algorithm can be 95.6%.The segmentation algorithm is with good accuracy and robustness. Also,the segmentation result is more in line with the visual characteristics of human.